homogenization
In plain wordsThe tendency of AI to make things more alike: products built on the same few models share the same flaws, and people who write or create with AI produce more similar work.
- 1
Foundation models
The convergence of many AI applications on the same few foundation modelsfoundation model, n.: A very large AI model, built once at great cost from vast amounts of text, images or code gathered mostly from the internet, that many products then reuse as their starting point, from chatbots to coding tools., so that whatever one model gets right or wrong is shared by everything adapted from it. The trend cuts both ways: a single improvement reaches every downstream product, but so does a single bias, blind spot or vulnerability, creating single points of failure. In decision-making, when many employers or lenders rely on the same model or data (algorithmic monoculture), the same people can be rejected everywhere, a pattern called outcome homogenization, and collective accuracy can fall even if each decision-maker's own accuracy rises.
Because all three hiring vendors used the same base model, a rejected applicant was rejected everywhere.
- 2
Culture and creativity
The narrowing of diversity in human outputs, ideas, style and language when many people use the same AI systems. Experiments find that writing with model assistance can make individual pieces more polished or rated more creative while making the collection of pieces more similar, and models tend to reproduce dominant cultural perspectives and a recognizable house style. Concerns extend to science, education and the diversity of online text available to train future models.
Origin
Rishi Bommasani and colleagues named homogenization, alongside emergence, as a defining feature of foundation models in 'On the Opportunities and Risks of Foundation Models' (Stanford, 2021); Jon Kleinberg and Manish Raghavan analyzed algorithmic monoculture the same year. Studies such as Anil Doshi and Oliver Hauser's (2024) documented the creative effect.
Related
- foundation modelfoundation model, n.: A very large AI model, built once at great cost from vast amounts of text, images or code gathered mostly from the internet, that many products then reuse as their starting point, from chatbots to coding tools.
- model collapsemodel collapse, n.: The gradual decline of AI models trained on the output of earlier AI models: rare details vanish first, then quality and variety erode, like a photocopy of a photocopy.
- filter bubblefilter bubble, n.: A situation in which you see mostly news and opinions you already agree with, because apps choose what to show you based on what you clicked before.
- AI slopAI slop, n.: Low-quality content churned out in bulk by AI — bland articles, fake-looking images, filler videos — posted to grab clicks or fill space, with little care for accuracy or value.
- LLM-ismsLLM-isms, n.: Words, phrases and habits of style that AI chatbots overuse — certain fancy verbs, tidy lists of three, 'it's not just X, it's Y' — which make writing recognizable as machine-made.
- concentration of powerconcentration of power, n.: The worry that AI will put enormous economic, political or military power in the hands of a few companies, governments or individuals, beyond the reach of ordinary checks and balances.
Sources
- Bommasani, R. et al. (2021). On the Opportunities and Risks of Foundation Models. arXiv:2108.07258.
- Kleinberg, J. & Raghavan, M. (2021). Algorithmic monoculture and social welfare. Proceedings of the National Academy of Sciences 118(22).
- Doshi, A. R. & Hauser, O. P. (2024). Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances 10(28).
Mentioned in
filter bubblefilter bubble, n.: A situation in which you see mostly news and opinions you already agree with, because apps choose what to show you based on what you clicked before., foundation modelfoundation model, n.: A very large AI model, built once at great cost from vast amounts of text, images or code gathered mostly from the internet, that many products then reuse as their starting point, from chatbots to coding tools.